Ej. Feleppa et al., TYPING OF PROSTATE TISSUE BY ULTRASONIC SPECTRUM ANALYSIS, IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 43(4), 1996, pp. 609-619
Prostate cancer is the highest-incidence cancer and second-leading can
cer killer of men in the U.S. Diagnosis now relies virtually exclusive
ly on core-needle biopsy, guided by transrectal ultrasound (TRUS). Bec
ause of the limitations of TRUS in detecting suspicious regions, biops
y often fails to sample cancer that is present or to determine that ex
tracapsular cancer exists, which results in false-negative biopsies or
inappropriate prostatectomies. Therefore, we conducted this study to
investigate the use of spectrum analysis of radio frequency (RF) echo
signals as a possible means of reducing the number of false-negative b
iopsies and inappropriate prostatectomies. This method utilizes databa
ses of parameters derived from normalized power spectra of RF echo sig
nals and histologically proven tissue types to determine ranges of par
ameter values associated with tissue types of interest. Typing an unkn
own tissue is performed by comparing the parameter values of the unkno
wn to the value ranges of specific tissue types In the database, Our r
esults provide encouraging preliminary discriminant-function distribut
ions that suggest an excellent potential for differentiating cancerous
from noncancerous prostate tissue far superior in terms of sensitivit
y and specificity than means now used to determine whether biopsy is r
equired. In addition, we developed images using color to indicate the
most likely tissue type throughout the tissue cross section as determi
ned by comparisons with database parameter values. These images showed
excellent correlation with histology.